# IDC automation rates

What automation rates are you getting?

• 0%-20%
• 20%-40%
• 40%-60%
• 60%-80%
• 80%-1000%
0 voters

The reason I’m asking this is while our automation rates have gone up, we seem to be stagnating now at about 18%. I’m using the automation rate reports and pivoting on when automation is 100% for true, and not for false.

This doesn’t mean that it’s not reading a lot of the fields, but to be able to have the system fully read the document without a user having to mess with the document is the goal here.

What are you guys getting? It 18% about as good as I can reasonably get?

1 Like

I am not sure, is there a reporting service to run that would tell me that?

I know it’s gotta be low.

there’s a reports tab

Then look at automation rates, data capture. Then export that.

Then in excel, run a pivot. (If ancora is watching this, these reports would be good to have…)

3 Likes

So, I’m looking at that dataset. The Document Automation rate is really only the number of documents (per batch) that processed correctly, versus those that had an Orange or Red flag. What does that really mean, statistically?

I would think that the “Fields read with Certainty” would be a better indication of the IDC success rate. I say that because any number of reasons outside of IDC’s control can cause the orange/red flags.

Additionally, because that rate column is “per batch” it will inherently skew any cumulative result if your batch size varies by more than n documents - meaning the average will skew based on the overall delta in batch size. eg - 100 batches of 2 documents versus 10 batches of 100 documents. If the smaller batches have a <50% automation rate, but the larger batches have 100% rate, what’s the true result you want to see?

(Anytime I see stats, I have to take it apart and question it - it’s my OCD)

I’m thinking that we need to use these columns to do the math.

NewCol = Source Fields from dataset
NetDocs = TotalDocs - Docs Deleted
GoodDocs = Complete(Green)
SuccessRate = NetDocs/GoodDocs

(will also need some date param and potentially separated by BatchType if more than one type is in the dataset)

I did this off the cuff, so I might be off-base.
Please comment - validate/agree/disagree with the math or logic

1 Like

Yeah, you’re right. We have as a rule of thumb one invoice per batch, so I’m using batch as a proxy to documents.

If you instead use the data accuracy tab instead, it’s per document. And 100% means that it read all of the fields correctly per document. And I am filtering this by a single document type. If you filter the grid before you export, it only exports what’s in the grid.

If I do the pivot on that, it’s a little better, at 20% for June.

1 Like

Yes - that would solve the fundamental problem.

But yeah, I’ve never really gotten into the reports - I should’ve looked for that report instead going off half-cocked. This dataset looks like a much better analysis of what IDC is really doing - good or bad.

My overall average is about the same - 89% on a batch type with 16 fields. usually always misses 2 fields.

How are you doing the true/false again?

True is where accuracy = 100. If it’s 100 it means that it can be read and processed with no user intervention.

right

1 Like

We grind the same vendors since we started and also do one invoice per batch and my idc is tweaked for those documents.

In 2024 we have done 61068 documents and 54660 were at 100 so 89.5%
2023 is similar with 111.336 with 100,068 at 100% of the 11,268 less than 100 8,280 were over 80% accurate.

1 Like

Greg killed it.

1 Like

You could always make an Epicor Idea to get this reviewed by Epicor ECM and then they could pass it on to the Ancora dev team.

2 Likes